The AI price war is intensifying, putting pressure on leading companies like OpenAI and Anthropic. Competitors are launching low-cost models to capture market share.
In the AI industry, a 'price war' is intensifying as model inference costs drop sharply. Following Meta's free release of Llama 3, several startups are offering GPT-4-level performance at much lower prices, shaking the price competitiveness of established leaders like OpenAI and Anthropic.
Historically, AI models commanded premium prices due to high development costs, but advances in open-source models and inference optimization have driven costs down. Meta's Llama 3 release, in particular, significantly boosted the performance of free models, narrowing the gap with commercial ones. As a result, enterprise customers are shifting to cheaper alternatives, pressuring OpenAI and Anthropic to lower prices.
This price war signals a fundamental shift in the AI industry's revenue model. Lowering prices to attract customers hurts profitability, while maintaining prices risks losing market share. In the long term, business models may evolve toward generating revenue from specialized services or platforms rather than the models themselves. Competition between big tech firms and startups is also expected to intensify.
Users compare AI model prices based on real experience, highlighting that DeepSeek offers performance similar to Claude at $20/month versus hundreds to thousands of dollars. Practical advice suggests choosing models based on task difficulty and that cheaper models require breaking tasks into smaller pieces. The full WSJ article was shared via a gift link due to the paywall.